Abstract
The energy storage batteries, employed in solar systems installed on lampposts, are usually placed in devices such as switchboards fixed at an elevation near the top of the column. However, this storage solution becomes inefficient, because it is not possible to guarantee the control of the working temperature of the batteries, due to the low thermal insulation capacity of these storage devices. In this sense, an underground compartment made of concrete, steel plate and rock wool were created, embedded in the foundation of the lamppost, with the purpose of using geothermal energy to maintain an adequate temperature inside the compartment. To verify the temperature inside the battery storage compartment, a thermal analysis was performed, where heat transfer by conduction, convection and radiation was considered. Analyses were performed in steady state, and later, transient state, considering the initial temperatures of the thermal study in the previous steady state. With a storage volume of 1m3 and the base of the compartment at a depth of 2m, it was verified that it is possible to use geothermal energy to cool or heat, depending on the season, a system through geothermal energy. Considering a typical day in July, with room temperature of 35oC, a reduction of approximately 8oC was obtained inside the storage compartment, compared to the ambient temperature.

Abstract
The article provides a comparison of Convolutional Neural Network (CNN) and Reinforcement Learning (RL) applied to the field of autonomous driving within the CARLA (CAr Learning to Act) simulator for training and evaluation. The analysis of results revealed CNNs better overall performance, as it demonstrated a more refined driving experience, shorter training durations, and a more straightforward learning curve and optimization process. However, it required data labelling. In contrast, RL relayed on an exhaustive (unsupervised) exploration of different models, ultimately selecting the model at timestep 600,000, which had the highest mean reward. Nevertheless, RL’s approach revealed its susceptibility to excessive oscillations and inconsistencies, necessitating additional optimization and tuning of hyperparameters and reward functions. This conclusion is further substantiated by a range of used performance metrics (objective and subjective), designed to assess the performance of each approach. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

Abstract
Sustainable development is crucial to ports due to the interconnection between port activities, the economy, and the environment. This study aims to explore how port digitalization initiatives played the role of promoting sustainable development. To this purpose, the author/authors adopted a mixed methods approach using as database the World Ports Sustainability Program, which features 74 port digitalization initiatives. The first step focused on a quantitative analysis of the distribution of said initiatives in terms of sustainable development goals, followed by a thematic analysis to explore their contribution. The findings indicate that more than 72% of ports addressed sustainable development goals 8, 9, 13, and 17. Digitalization initiatives in ports have mainly focused on improving their infrastructure and operational performance, enabling them to address climate change challenges. This work also recognized the role that partnerships can play in achieving this goal.

Abstract
When people move around a town, at some point in their journey they need to cross the road using a dedicated crosswalk. However, crossing is not always done safely due to weather conditions, lack of visibility or distraction. The VALLPASS project, aims to install two lampposts in opposite positions to the direction of crossing, with various functionalities and technological innovations, creating a luminous tunnel for the safe passage of pedestrians. To verify the mechanical resistance of the lighting poles, numerical simulations were performed using the finite element method, where the boundary conditions considered the criteria defined by the European standard EN-40 "Lighting Columns". This standard specifies the loads acting on the column, namely the horizontal forces due to the action of wind according to standard NP EN 1991-1-4:2010 and the vertical forces due to the self-weight of the entire structure. Considering a lighting pole with a square lower section and a cylindrical upper section, with a total height of 7 meters and with a support structure for photovoltaic panels, according to the static analysis performed, a maximum combination of axial and bending stresses of 138.74MPa, was obtained in the connection zone between the square section and the pole shaft. The maximum displacement of 6.9cm, was obtained at the free ends of the photovoltaic panel support structure and a minimum factor of safety of 1.64 in the zone where the combination of axial and bending stresses is more severe.

Abstract
The increased interest and importance of explaining neural networks' predictions, especially in the medical community, associated with the known unreliability of saliency maps, the most common explainability method, has sparked research into other types of explanations. Natural Language Explanations (NLEs) emerge as an alternative, with the advantage of being inherently understandable by humans and the standard way that radiologists explain their diagnoses. We extend upon previous work on NLE generation for multi-label chest X-ray diagnosis by replacing the traditional decoder-only NLE generator with an encoder-decoder architecture. This constitutes a first step towards Reinforcement Learning-free adversarial generation of NLEs when no (or few) ground-truth NLEs are available for training, since the generation is done in the continuous encoder latent space, instead of in the discrete decoder output space. However, in the current scenario, large amounts of annotated examples are still required, which are especially costly to obtain in the medical domain, given that they need to be provided by clinicians. Thus, we explore how the recent developments in Parameter-Efficient Fine-Tuning (PEFT) can be leveraged for this usecase. We compare different PEFT methods and find that integrating the visual information into the NLE generator layers instead of only at the input achieves the best results, even outperforming the fully fine-tuned encoder-decoder-based model, while only training 12% of the model parameters. Additionally, we empirically demonstrate the viability of supervising the NLE generation process on the encoder latent space, thus laying the foundation for RL-free adversarial training in low ground-truth NLE availability regimes. The code is publicly available at https://github.com/icrto/peft-nles.

Abstract
This study focuses on the analysis of emulsion pictures to understand important parameters. While droplet size is a key parameter in emulsion science, manual procedures have been the traditional approach for its determination. Here we introduced the application of YOLOv7, a recently launched deep-learning model, for classifying emulsion droplets. A comparison was made between the two methods for calculating droplet size distribution. One of the methods, combined with YOLOv7, achieved 97.26% accuracy. These results highlight the potential of sophisticated image-processing techniques, particularly deep learning, in chemistry-related topics. The study anticipates further exploration of deep learning tools in other chemistry-related fields, emphasizing their potential for achieving satisfactory performance.